Abstract WP45: VISIION-S: Viz.ai Implementation Of Stroke Augmented Intelligence And Communications Platform To Improve Indicators And Outcomes For A Comprehensive Stroke Center And Network - Sustainability

Background: As Comprehensive Stroke Centers (CSCs) strive to improve neurointerventional (NIR) times, process improvements have been put in place to streamline workflows. Our prior publication (VISIION) demonstrated an improvement in key performance indicators (KPIs) in our CSC. The purpose of this study is to analyze whether the positive results demonstrated were sustainable.

Methods: Sequential stroke NIR patients being Direct Arriving LVO (DALVO) and telemedicine transfer LVO (BEMI) cases were assessed, including subgroups of DALVO-OnHours, DALVO-OffHours, BEMI-OnHours, and BEMI-OffHours. We analyzed times for the original 6 months pre (6/10/20-1/15/21) and compared them to a 17 months post-implementation (1/16/21- 6/25/22) to evaluate for sustainability. Mann-Whitney U was utilized.

Results: 150 NIR cases were analyzed pre (n=47) v. post (n=103) Viz.ai implementation (DALVO-OnHours 7 v. 20, DALVO-OffHours 10 v. 25, BEMI-OnHours 13 v. 20, BEMI-OffHours 17 v. 38). For Door-to-groin (DTG) assessments, improvement was noted for DALVO-OffHours 39% (157min,96min;p<0.001), DALVO-ALL 25% (127min,95min;p=0.006), BEMI-OffHours 46% (45min,25min;p=0.023), and BEMI-ALL 40% (42min,25min;p=0.005). Activation-to-groin (ATG), door-to-device (DTD), and door-to-recanalization (DTR) showed similar improvements. For DALVO-OffHours, there were significant reductions in door to CT (DTC) 81% (26min,5min;p<0.001), ATG 32% (90min,61min;p=0.036), DTG 39% (157min,96min;p<0.001), DTD 31% (178min,123min;p=0.002), and DTR 32% (197min,135min;p=0.003).

Conclusions: Consistent with our initial 6 month post-implementation pilot, we noted sustainability over a 17 month period with sustained reduction in KPIs for numerous key NIR subgroups. In the greatest opportunity subgroup (DALVO-OffHours), requiring team mobilization off hours without benefit of telemedicine transfer lead time, we noted a significant reduction in all 5 time metrics. Our sustainability finding is important to show that process improvements continued even after the immediate period, making a Hawthorne effect less likely and adding credibility to the results. Models such as this, could be useful for other centers striving to optimize workflow and improve NIR times.

Abstract WP84: (VISIION): Viz.ai Implementation Of Stroke Augmented Intelligence And Communications Platform To Improve Indicators And Outcomes For A Comprehensive Stroke Center And Network

Background: Comprehensive Stroke Centers (CSCs) strive to narrow rt-PA and “Door To Groin” (DTG) neurointerventional (NIR) times. Process improvement workflows have been put in place for rt-PA. While similar processes have been implemented to streamline workflows for hyperacute NIR cases, complex pathways, disparate imaging locations, and fragmented communications all highlight a need for continued improvements.

Methods: This quality improvement initiative (IRB #210525) was implemented to assess our transition to the Viz.ai platform for immediate image review and centralized communication and its effect on key performance indicators (KPIs) in an already robust CSC. We compared 6 month periods prior to and following deployment. Sequential stroke NIR patients were included. Both Direct Arriving LVO (DALVO) and telemedicine transfer LVO (BEMI) cases were assessed. We assessed subgroups of DALVO-OnHours, DALVO-OffHours, BEMI-OnHours, and BEMI-OffHours. Mann-Whitney U was utilized.

Results: Eighty-two NIR cases were analyzed pre v. post Viz.ai implementation (DALVO-OnHours 7 v. 7, DALVO-OffHours 10 v. 5, BEMI-OnHours 13 v. 6, BEMI-OffHours 17 v. 17). DALVO-OnHours improved 19% (97min, 79min; p=0.201) in median DTG times. DALVO-OffHours had a significant 39% reduction (157min, 95min; p=0.009). DALVO-“All” showed a significant 32% reduction (127min, 86 min; p=0.006). BEMI-OnHours improved 18% (37min, 31min; p=0.337). BEMI-OffHours improved 38% (45min, 28min; p=0.077). BEMI-“All” significantly improved 33% (42min, 28min; p=0.036). Overall, there was a 22% reduction (50min, 39min; p=0.066) after Viz.ai implementation.

Conclusions: There was an immediate KPI improvement following Viz.ai implementation for both direct arrival and telemedicine transfer NIR cases (32% and 33% respectively). In the greatest opportunity subset (direct arriving cases requiring team mobilization off hours without benefit of telemedicine transfer lead time) we noted a 39% improvement. With Viz.ai, we noted immediate access to images and streamlined group communications, even in an already well-functioning CSC. These results have implications for future care processes and can be a model for centers striving to optimize workflow and improve NIR timeliness.

Abstract MP23: Clinical and Demographic Factors Influence Clinical Trial Enrollment

Introduction: Inclusion of diverse patients in clinical trials is essential to represent patient populations and to ensure that results are generalizable. Analysis of this topic is motivated by prior reports of clinical trials underrepresenting historically disadvantaged groups. We aimed to understand if the clinical trial population is representative of the overall stroke population in our community.

Methods: We obtained clinical and demographic data from hospital administrative records including all acute ischemic stroke patients between March 2019 and February 2020. Consent was counted when a signed consent for study participation was documented in the inpatient record. Enrollment in studies with a primary inpatient population at the time of consent (MOST, TIMELESS, ARCADIA, SleepSMART, and STRONG) were included. Patients with consent (CP) and without consent (NCP) were compared using chi-square analysis and t-test probability with SPSS.

Results: During the study period, 504 patients met the above criteria; 55 were consented to participate in clinical trials. Overall CP did not differ from NCP in % women (45.5% vs 41.2%, not significant), Hispanic ethnicity, discharge medications, hypertension, diabetes, heart failure, drug and alcohol abuse, atrial fibrillation, and payer (Medicare versus non-Medicare). Median age was lower in CP compared to NCP (63 ± 12.8 vs. 67 ± 13.9, p=.046). The mean NIHSS was lower in CP (6.37 ± 5.95 vs. 8.21 ± 8.91, p=.048). The CP group had lower incidence of prior stroke (9.1% vs. 20.9%, p=.037), less dyslipidemia (12.7% vs. 30.1%, p=.007), fewer comorbidities on average (1.9 ± 1.1 vs. 2.2 ± 1.5, p=.044), was less likely to identify as Asian (0% vs. 8.4%, p=.015) and more likely to identify as white (59.3% vs. 45.7%, p=.045). The CP group was more likely to have migraine (12.7% vs. 1.8% p<.001) and were discharged on more stroke risk-factor modifying medications (2.2 ± .558 vs. 1.9 ± .902, p=.001).

Conclusions: While the study population is largely representative, they are more likely to be White, with lower stroke severity, fewer medical comorbidities and less likely to be Asian. We plan to expand this analysis to include more study centers and to guide future clinical trial design.

Abstract P724: Perimenopausal Women With Migraine Present With Stroke at a Younger Age and With Less Comorbid Diabetes

Introduction: Migraine is an independent risk factor for ischemic stroke. Frequency and severity increases in the perimenopausal period accompanied by marked vasomotor symptoms (VMS) such as hot flashes, flushing and night sweats. There is emerging evidence that VMS increases the risk of vascular disease including stroke. The purpose of this study was describe the demographics and co-morbidities of perimenopausal females with and without migraine that experience acute ischemic stroke (AIS).

Methods: In this IRB approved study, electronic health record (EHR) data was obtained from a large, academic, comprehensive stroke center from 1/1/2015 to 1/1/2020. Inclusion criteria included female sex, age 42-65 years, and hospital diagnosis code of AIS. Hemorrhagic stroke, TIA, vasculopathy, and endocarditis associated strokes were excluded. Perimenopause was defined as age ≥42 and ≤65 years. Hormonal and menopausal status was not available in the EHR. We compared the baseline demographics and co-morbidities by ICD10 codes of subjects with and without migraine. Chi squared was used to compare categorical data and t test for continuous. Spearman rho was used to assess correlations.

Results: We identified 660 subjects who met study criteria (n=83 with migraine; n=577 without migraine). Migraine positive subjects were significantly younger (mean age 58 vs 66 years, p=0.03) at time of AIS. Migraine positive subjects identified significantly more often as White (47%) compared to Black (10%), Asian (7%), Pacific Islander (1%), Native American/Alaskan (1%), Other/Mixed Race (31%), and unknown (3%), p=0.001. There was no significant difference in Hispanic ethnicity (p=0.87), hypertension (p=0.66), hyperlipidemia (p=0.12), or atrial fibrillation (p=0.84). Comorbid diabetes was significantly higher in the non-migraine group (94% vs 6%, p<0.001).

Conclusion: Perimenopausal women with concomitant history of migraine present with AIS at younger ages and with lower rates of diabetes than those without a migraine history. Future research must be done to assess the correlation of menopausal symptom severity, hormone levels at time of AIS, and stroke characteristics to further understand the role of menopause in stroke risk.

Abstract P697: Vasomotor Symptoms of Menopause Increase Stroke Risk in Migraineurs

Introduction: The relationship between migraine and stroke, especially in migraine with aura, has been well-established. There is emerging evidence that vasomotor symptoms (VMS) such as hot flashes, flushing and night sweats associated with menopause increases the risk of vascular events, especially in the perimenopausal period. The aim of this study was to compare vascular risk factors in perimenopausal females with and without migraine with concomitant acute ischemic stroke (AIS).

Methods: In this IRB approved study, we examined patient level data using the SlicerDicer function within Epic at a large, academic, comprehensive stroke center from 1/1/2015 to 1/1/2020. Inclusion criteria included female sex, age 42-65 years, and hospital diagnosis code of AIS. Hemorrhagic stroke, TIA, vasculopathy, and endocarditis associated strokes were excluded. Perimenopausal was defined as age ≥42 and ≤65 years. Hormonal and menopausal status was not available. We compared rates of co-morbidities by ICD10 codes of subjects with and without migraine using descriptive statistics and Chi squared analysis.

Results: We identified 2296 (90%) women without migraine (Group 1) and 243 (10%) with migraine (Group 2) admitted for AIS. The five most common risk factors for AIS in group 1 were hypertension (56%), hyperlipidemia (37%), diabetes (30%), obesity (23%) and atrial fibrillation (11%). VMS was coded in 8% and tobacco use 7%. In group 2 we found hypertension (50%), hyperlipidemia (42%), migraine with aura (31%), obesity (23%), diabetes (20%). VMS was coded in 14%, atrial fibrillation in 12%, and tobacco use in 6%. Group 2 patients were more likely to have VMS (p = 0.008) and less likely to have diabetes (p=0.001). There were no other significant differences identified.

Conclusions: Vasomotor symptoms in menopause are a significant risk factor for AIS in perimenopausal women with migraine. VMS should be assessed for clinically and included as a risk factor for stroke, especially in those with additional vascular risk factors. Future studies should include a diverse sample to assess the impact of VMS in a heterogeneous population.

Abstract P698: Coming Together in a Time of Distancing: Creating Community Messaging for Emergency Care During Covid

Introduction: In the early months of COVID-19 pandemic, a decline in stroke hospital admissions were reported nationwide. In a large, diverse region of Southern California, a collaborative effort was made to collect real-time data trends in stroke code activations and to assess this impact locally. The San Diego (SD) County Stroke Receiving Centers demonstrated a notable decrease of 30% in stroke code activations from March-May 2020 as compared to the same timeframe in 2019, which motivated the group to dedicate time and resources to pursue a united community messaging focused on seeking emergency treatment for stroke.

Methods: A unified marketing campaign was created in collaboration with SD County EMS and the SD region American Heart Association/American Stroke Association. A single graphic message was utilized that emphasized the importance of seeking emergency treatment when suffering signs of stroke, along with the slogan “We are here for you. Every minute matters.” Impact of the campaign was gauged by quantifying the number of times our message was viewed on social media and number of stroke code activations after the campaign ended.

Results: The unified social media campaign was posted by 14 of the 18 SD County stroke receiving hospitals during the month of June 2020. The team utilized Facebook, Twitter, Instagram and LinkedIn to convey the message. The campaign yielded a total of 26,727 views. The median monthly stroke code activations in July 2020 increased to 34, as compared to 26.5 for March-May 2020.

Conclusion: In a time when social distancing has become the norm, it is more important than ever to band together as a community. This endeavor demonstrates that virtual messaging serves as a viable option for community education during the COVID-19 pandemic and in the future. A unified social messaging campaign targeting the importance of seeking emergency care for stroke during the COVID-19 pandemic is an effective way to reach large numbers of people regionally.

Abstract P666: Covid-19 Geographic Distribution and Stroke Code Activation Within San Diego County

Introduction: In the early months of the COVID-19 pandemic, decreased numbers of stroke code activations were reported nationwide. In San Diego County, a diverse region that borders Mexico with over 4500 square miles and population 3.3 million, trends in COVID-19 cases varied geographically. We saw an overall decrease in stroke cases across our systems and aimed to better understand if high COVID infection rates in subregions affected stroke code activations.

Methods: Stroke code activation data from 15 Stroke Receiving Centers were matched with COVID-19 case rates by patient home zip code. Patients arriving via emergency medical services (EMS) or private transportation were included. Patients with home zip codes outside of San Diego County were excluded. Data represented the cumulative rate of stroke codes and COVID-19 cases per 100,000 population per zip code for the period of March 1 through June 30, 2020.

Results: We counted 1,927 stroke code activations across 106 zip codes in San Diego County. The average stroke code activation rate was 58.4 per 100,000 (range: 0-310.6) The median stroke code activation rate was 55.95 (IQR=32.1-73.1) per 100,000 population. The median COVID rate per zip code was 244.9 (IQR=177-448.4) per 100,000 population. There were 958 (49.7%) non-stroke diagnoses, 576 (29.9%) AIS, 272 (14.1%) TIA, 104 (5.4%) ICH and 17 (.9%) SAH. We did not identify a correlation between stroke code activation rates and COVID rates across zip codes (r=.17, p=.09, 95% CI(-.02, .35)).

Conclusions: Across a large and diverse single-county region, no correlation was found between COVID positivity rate per zip code and stroke code activations. We found no decreases in stroke code activations in areas with high COVID rates.

A Stroke Care Model at an Academic, Comprehensive Stroke Center During the 2020 COVID-19 Pandemic

BACKGROUND AND PURPOSE

The COVID-19 pandemic has required the adaptation of hyperacute stroke care (including stroke code pathways) and hospital stroke management. There remains a need to provide rapid and comprehensive assessment to acute stroke patients while reducing the risk of COVID-19 exposure, protecting healthcare providers, and preserving personal protective equipment (PPE) supplies. While the COVID infection is typically not a primary cerebrovascular condition, the downstream effects of this pandemic force adjustments to stroke care pathways to maintain optimal stroke patient outcomes.

METHODS

The University of California San Diego (UCSD) Health System encompasses two academic, Comprehensive Stroke Centers (CSCs). The UCSD Stroke Center reviewed the national COVID-19 crisis and implications on stroke care. All current resources for stroke care were identified and adapted to include COVID-19 screening. The adjusted model focused on comprehensive and rapid acute stroke treatment, reduction of exposure to the healthcare team, and preservation of PPE.

AIMS

The adjusted pathways implement telestroke assessments as a specific option for all inpatient and outpatient encounters and accounts for when telemedicine systems are not available or functional. COVID screening is done on all stroke patients. We outline a model of hyperacute stroke evaluation in an adapted stroke code protocol and novel methods of stroke patient management.

CONCLUSIONS

The overall goal of the model is to preserve patient access and outcomes while decreasing potential COVID-19 exposure to patients and healthcare providers. This model also serves to reduce the use of vital PPE. It is critical that stroke providers share best practices via academic and vetted social media platforms for rapid dissemination of tools and care models during the COVID-19 crisis.

Abstract NS8: Increasing Discharges to Acute Rehabilitation With a Multidisciplinary Task Force

Background: Patients with Acute Ischemic Stroke recover function better when discharged to inpatient rehabilitation (IRF) over skilled nursing facilities (SNF). We aimed to increase discharge to IRF over SNF at two Comprehensive Stroke Centers (CSC) within our care network.

Methods: We reviewed the discharge pattern at our CSCs compared to comparable centers using GWTG benchmarking and designated a multidisciplinary task force aiming to meet discharge patterns at our centers to nationwide trends. The task force used Lean Methodology to identify barriers for IRF discharge recognition and placement.

Results: The taskforce identified non-modifiable barriers such as socioeconomic determinants, insurance status, family/social support; and modifiable barriers such as inconsistencies in therapist recommendations and notations, limited access to case management, and lack of provider knowledge about IRF admission criteria. Beginning November 1, 2018, the following interventions were used to achieve increased IRF referral and admission rates: educating therapists to provide more specific and consistent documentation; thorough therapist, case management, and provider education on IRF admission criteria; cohorting patients on dedicated neurology units; and daily multidisciplinary team meetings (consisting of a therapist, case management, and the primary provider) on the neurology units. IRF admission rates were then collected retrospectively. Between Novembers 1, 2018 to April 30, 2019, the rate of admissions to IRF for the UCSD Health System increased from 7.7% to 13.5% (see Table).

Conclusions: Using Lean Methodology we identified and reduced barriers for IRF referral after stroke. This suggests IRFs are underutilized when disposition is not effectively streamlined. Further studies are needed to understand which interventions had the highest impact of increasing IRF admission and referral rate.